Enterprise IT vendors and consultancies regularly publish industry surveys as part of their marketing efforts. The polls that focus on AI usually ask executives how their companies are implementing the technology. Capgemini, the French consulting heavyweight, took a different route with one of its newest reports. Instead of polling executives about their companies’ automation programs, it asked how they’re using AI in their own work. The firm collected responses from 500 C-Suite members.

Capgemini surveyed, among others, business leaders whom it defines as active AI users. The vast majority of executives who fall into that category use AI at least daily. The report’s first notable finding is that this segment is growing rapidly. According to Capgemini, the percentage of C-Suite leaders who are active AI users will more than double to 38% within 3 years.

Not too surprisingly, technical executives are adopting AI more extensively than their colleagues. Currently, 26% of them use the software actively. Capgemini expects that number to reach 53% in 3 years, which will put technical executives about 15% ahead of the broader C-Suite. CEOs, who made up a fifth of the survey’s respondents, are the second most enthusiastic adopters of AI. More than a third of them will use tools such as chatbots actively in 2029.

Capgemini also looked into what tasks executives are automating with AI. Today, those tasks mainly relate to managing emails and online research. The consultancy expects that to change significantly in the next few years. According to Capgemini, the subset of C-Suite executives who use AI to critique or simulate business moves will jump more than sixfold to 38% within 3 years. A small percentage of respondents indicated that they will take it a step further. Those executives, who made up about 6% of the survey group, said that they plan to use the technology to generate “semi-autonomous decisions under human supervision.”

Capgemini released the study alongside a second report that looked at organizational-level AI initiatives. The latter survey drew on a broader respondent pool that comprised 1,505 executives. According to Capgemini, the participants said that their companies plan to boost AI spending from 3% of their budgets in 2025 to 5% this year.

At the same time, enterprises are becoming more deliberate about where they invest. About two thirds of the respondents said that their companies plan to “rationalize or consolidate” their AI initiatives this year to prioritize high-impact projects. A similar percentage said that they have already started pausing “lower-value” investments.

One way companies are driving down AI costs is by switching to more resource-efficient LLMs. Proprietary foundation models are usually delivered via the cloud and billed based on usage, which means that they can be expensive to use at scale. Open-source LLMs are inherently more economic. According to Capgemini, 64% of the respondents to its survey said that their companies are exploring either customized versions of open-source LLMs or SLMs. Those are small, hardware-efficient language models that incur lower inference costs than a standard LLM.

Sales and marketing are the two areas that companies are prioritizing most heavily with their AI investments. According to Capgemini, 63% of the respondents listed those fields as the biggest focus of their firms’ AI programs. IT operations followed in second place with 55%. At the technical level, meanwhile, data foundations and pipelines are expected to be the fastest-growing AI line item for IT apartments. Data pipelines are the software workflows that companies use to prepare information for use in LLM training. Compute infrastructure took second place, while model development and licensing ranked third. 

Photo courtesy of Capgemini